Neurotransmitter release is mediated by the fusion of synaptic vesicles with the presynaptic plasma membrane. Fusion is triggered by a rise in the intracellular calcium concentration and is dependent on the neuronal SNARE (soluble N-ethylmaleimide-sensitive fusion protein-attachment protein receptor) complex. A plethora of molecules such as members of the MUNC13, MUNC18, complexin and synaptotagmin families act along with the SNARE complex to enable calcium-regulated synaptic vesicle exocytosis. The synaptotagmins are localized to synaptic vesicles by an N-terminal transmembrane domain and contain two cytoplasmic C2 domains. Members of the synaptotagmin family are thought to translate the rise in intracellular calcium concentration into synaptic vesicle fusion. The C2 domains of synaptotagmin-1 bind membranes in a calcium-dependent manner and in response induce a high degree of membrane curvature, which is required for its ability to trigger membrane fusion in vitro and in vivo. Furthermore, members of the soluble DOC2 (double-C2 domain) protein family have similar properties. Taken together, these results suggest that C2 domain proteins such as the synaptotagmins and DOC2s promote membrane fusion by the induction of membrane curvature in the vicinity of the SNARE complex. Given the widespread expression of C2 domain proteins in secretory cells, it is proposed that promotion of SNARE-dependent membrane fusion by the induction of membrane curvature is a widespread phenomenon.
At the chemical synapse, communication occurs by the regulated release of neurotransmitters from presynaptic neurons into the synaptic cleft where they can activate receptors on the postsynaptic plasma membrane. In the presynaptic terminal, the neurotransmitters are stored in small, 40–50 nm diameter vesicles. Release of neurotransmitters is achieved by the fusion of these synaptic vesicles with the plasma membrane and is triggered by an increase in the calcium concentration in the presynaptic terminal .
The hemifusion model of membrane fusion
The fusion of synaptic vesicles with the plasma membrane entails the merger of two initially separated membranes. Membrane fusion processes are ubiquitous, being essential for many aspects of eukaryotic cell biology. Thus intracellular membrane traffic, the fusion of mitochondria or the infection of host cells by enveloped viruses require membrane fusion . Over the last years, a consensus model about the mechanism by which membranes fuse has emerged  (Figure 1). According to this model fusion is initiated by the local destabilization of the two membranes destined to fuse. Destabilization favours the transition into a so-called hemifusion stalk in which the two contacting but not the two distal monolayers are merged. Hemifusion is followed by the merger of the two distal monolayers resulting in the opening of the fusion pore. This is the first time contents of the two membrane enclosed compartments mix. At the synapse the neurotransmitters would be in contact with the extracellular space. After fusion pore opening, the fusion pore may enlarge at the synapse, leading to full collapse of the synaptic vesicle into the plasma membrane.
The hemifusion model of membrane fusion
For membrane fusion to occur, several energy barriers have to be overcome. Somewhat counter intuitively, it has now been established that the barriers for fusion become progressively larger during the fusion process. Thus the formation of the hemifusion intermediate requires less energy than the opening of the fusion pore, which in turn is energetically more favourable than fusion pore dilation .
During membrane fusion, the main conformational changes are undergone by the merging membranes, which are regulated by cellular proteins. For synaptic vesicle fusion the task for the presynaptic fusion machinery is to rapidly decrease the energy barriers for fusion only on an increase in the intracellular calcium concentration. A plethora of proteins has been identified that are required for the exocytosis of synaptic vesicles. Among them are members of the MUNC13, MUNC18, complexin, synaptotagmin and SNARE (soluble N-ethylmaleimide-sensitive fusion protein-attachment protein receptor) protein families . The main task is now to assign specific functions to these proteins and to unravel their precise mechanism of action.
The SNARE complex
Probably the most prominent and best studied of the exocytic proteins are the SNAREs. The SNAREs required for synaptic vesicle fusion are synaptobrevin, which is localized to synaptic vesicles by a C-terminal transmembrane domain and the two plasma membrane SNAREs syntaxin1 and SNAP-25 (25 kDa synaptosome-associated protein). Syntaxin1 contains a C-terminal transmembrane domain, whereas SNAP-25 is targeted to the membrane by four palmitoylated cysteine residues. The three SNAREs have the ability to form an extremely stable four-helical bundle to which synaptobrevin2 and syntaxin1 contribute one helix each and SNAP-25 two helices [5,6]. This four-helical bundle is also referred to as the SNARE complex. SNARE complex formation is thought to bring the plasma and vesicular membranes into close proximity, aiding membrane fusion . Indeed, SNARE complex formation has been shown to be essential for synaptic vesicle fusion and for a variety of other intracellular membrane fusion events . Moreover, SNAREs can mediate membrane fusion in reconstituted systems with varying efficiency depending on the protein density, the curvature and lipid composition of the reconstituted vesicles and other experimental conditions [8–12]. It has, however, also become increasingly clear that, at physiological protein densities, SNAREs are rather poor fusogens and that, at higher densities, SNARE-mediated fusion is accompanied by extensive lysis [12,13]. Also, mechanistically it is not clear how the force generated on SNARE complex formation can be transferred into the membrane as the short linker between the SNARE domain and the transmembrane domain appears to be flexible [2,11,14,15].
As mentioned above, SNAREs do not act alone during synaptic vesicle exocytosis (and indeed during any other SNARE-dependent membrane fusion event investigated so far). Among the best-studied molecules acting alongside the SNARE complex are the synaptotagmins and, in particular, synaptotagmin-1. Synaptotagmin-1 is localized to synaptic vesicles by an N-terminal transmembrane domain. The main functional domains of synaptotagmins are two cytoplasmic C2 domains, which show calcium-dependent SNARE and membrane binding [16,17]. Synaptotagmin-1 is a member of a protein family with 17 members in humans. Several of these, in particular synaptotagmin-2, -3, -5, -7 and -9, show calcium-dependent membrane and SNARE binding . Synaptotagmin-1 has gained prominence since it has been shown to translate the rise in the intracellular calcium concentration into synaptic vesicle exocytosis. In particular, studies on synaptotagmin-1-knockout mice have shown that synaptotagmin-1 is required for the synchronous phase of neurotransmitter release . Synchronous neurotransmitter release is tightly coupled with the action potential-triggered opening of voltage-dependent calcium channels. In contrast, asynchronous neurotransmitter release, occurring with a slight delay after calcium channel opening, and spontaneous release, occurring in the absence of an action potential, are independent of synaptotagmin-1. The calcium sensors for asynchronous and spontaneous synaptic vesicle fusions are unknown.
A number of studies have shown that both the calcium-dependent membrane-binding and SNARE-binding activities of synaptotagmin-1 are required to trigger fusion. Thus mutations in synaptotagmin-1 that reduce the affinity for membranes also reduce the probability of synaptic vesicle fusion [19–21]. In contrast, mutations that increase the affinity of synaptotagmin-1 for membranes result in an increased probability of synaptic vesicle fusion . On the other hand, synaptotagmin-1 has been shown to bind the components of the SNARE complex in a calcium-dependent manner [16,20,23–25] and mutations that reduce this interaction result in a loss of fusion triggering by synaptotagmin-1 [20,24,26,27]. In addition, synaptotagmin-1 shows calcium-independent interactions with SNAREs and the plasma membrane phospholipid PIP2 (phosphatidylinositol 4,5-bisphosphate) [28,29].
We have recently shown that, on calcium-dependent membrane binding and insertion by its C2 domains, synaptotagmin-1 induces a high degree of membrane curvature . We have further shown that membrane curvature induction by synaptotagmin-1 is required for the promotion of SNARE-dependent fusion in vitro . Furthermore, experiments in PC12 cells have shown that membrane curvature induction by synaptotagmin-1 is also required for the calcium-dependent exocytosis of dense core granules . The PC12 cell experiments further revealed that synaptotagmin-1-induced membrane curvature promotes opening of the fusion pore, indicating that synaptotagmin-1 has a role during all stages of fusion (Figure 1).
DOC2 (double-C2 domain) proteins
Evidence has shown that proteins of the DOC2 family are in many ways analogous to the calcium-dependent synaptotagmins such as synaptotagmin-1 in terms of membrane- and SNARE-binding and membrane curvature induction. DOC2A and DOC2B have, like the synaptotagmins, two C-terminal C2 domains but, unlike the synaptotagmins, have no N-terminal transmembrane domain and are thus soluble. Instead they contain an N-terminal MUNC13-interacting domain. DOC2 proteins have been shown in many different cellular systems to be calcium-dependent pro-exocytic proteins. Thus overexpression of DOC2B in chromaffin cells increases calcium-dependent granule exocytosis . Moreover, the overexpression of DOC2B has a direct effect on the behaviour of the fusion pore, promoting its expansion . This result indicates that DOC2B is an integral part of the fusion apparatus. In addition, DOC2 proteins have been shown to promote exocytosis in other cell types such as β-cells , mast cells  and PC12 cells .
Based on these results, the following model for synaptotagmin-1 and other calcium-dependent C2 domain proteins such as DOC2B is proposed [2,27,30]. Calcium binding by the two C2 domains results in the interaction with and insertion into the plasma membrane [35,36]. It is possible that C2 domains bind to the vesicular membrane but given its higher net-negative charge, we consider the plasma membrane a likelier target for the C2 domains. As a result of membrane insertion, the C2 domains of synaptotagmin-1 buckle the plasma membrane towards the vesicle (Figure 2). This has two major effects. First, the distance between the plasma and vesicular membrane is reduced and, secondly, the lipids in the protein-free end cap of the buckled membrane are under curvature stress. This curvature stress is released during the fusion process. Both the decrease of the inter-membrane distance and the curvature stress dramatically increase the probability of fusion. Since membrane binding by synaptotagmin-1 is calcium-dependent, this model provides an explanation for the calcium-dependent triggering of synaptic vesicle fusion by synaptotagmin-1. In its simplest form, SNARE complex binding by synaptotagmin-1 merely serves to target the membrane curvature to the site where the vesicle is tethered to the plasma membrane. However, it is also possible that synaptotagmin-1 binding to the SNAREs directly affects SNARE complex assembly and/or release of the complexin fusion clamp [16,37–40]. Thus synaptotagmin-1, and by analogy other DOC2 proteins, have a central role during membrane fusion acting as fusogens alongside the SNARE complex.
Model of synaptotagmin-1-triggered fusion initiation
Proteins with multiple linked C2 domains are widely expressed and it is therefore possible that the promotion of SNARE-dependent membrane fusion by the C2 domain-driven induction of membrane curvature is a widespread phenomenon.
Molecular Mechanisms in Exocytosis and Endocytosis: 7th Junior Academics Meeting, an Independent Meeting held at University of Edinburgh, Edinburgh, U.K., 5–7 April 2009. Organized and Edited by Rolly Wiegand (Edinburgh, U.K.).
I am grateful to Dr Harvey McMahon and Dr Rohit Mittal for critically reading the manuscript.
The European Molecular Biology Organization [long-term fellowship LTF 21-2006] and the Medical Research Council are gratefully acknowledged for funding this work.